Research And Implementation Of Intelligent Transportation Customer Service System Based On Domain Knowledge Base

With the rapid development of the transportation industry and Internet technology,information data is increasing day by day,and people’s life rhythm is accelerating.In the field of transportation customer service,the training and management costs of manual customer service are relatively high and cannot meet the needs of users for real-time consultation.The traditional customer service system distributes different information under various department information portals.Variety will lead to very low efficiency,this model has been unable to meet the increasing demand of the majority of driver users to efficiently obtain information,so the research and implementation of intelligent transportation customer service system is urgent.This thesis designs and implements an intelligent transportation customer service system with high accuracy and very stable performance.The system uses WeChat public account as the platform,which not only solves the problem of slow loading in the past when developing systems on webpages,but also unified user queries the entrance of information can make it easier to summarize the questions raised by users.After analyzing the research on the intelligent question answering system at home and abroad,this thesis mainly improves on the three aspects of domain knowledge base design,intention recognition,and system design.The core technology is studied.The main contributions and results of this thesis are as follows:(1)This thesis analyzes the characteristics of the existing four types of intelligent question answering systems and finds that these intelligent question answering systems cannot independently help us solve problems.After combining the needs of users in the field of transportation customer service,a new type of question-answer storage model is designed.,A method that combines domain knowledge base and frequently asked questions(Frequently Asked Questions,FAQ)is proposed and applied to intelligent transportation customer service system.The design idea of the domain knowledge base is as follows: first,the questions frequently asked by users are divided into domains and intents according to business logic,and then the answer pairs of the questions are expressed in the form of triples,that is,the form of "entity-relationship(answer)-entity",The answer here is assigned to the relationship between two entities in the form of attributes,and finally the question-answer pairs are stored in the relational database in the form of triples.This method effectively solves the problems of the high cost of manually constructing knowledge base in the traffic customer service question answering system and high requirements on question answering efficiency and accuracy.(2)In this thesis,the intent recognition technology is improved.The intent recognition task is usually regarded as a classification task and the problem of user consultation is usually short text.Therefore,it is proposed to use BERT(Bidirectional Encoder Representation from Transformers)in combination with CNN(Convolutional Neural Networks)model Identify.This method achieves the best performance on both Chinese and English datasets.In this thesis,the main operation steps of the model for intent recognition using BERT and CNN include preprocessing the short text,then putting the processed short text into the model to obtain a vectorized representation of the short text,and finally identifying it by the intent classifier The intention of short text,short text preprocessing includes text cleaning,stop word removal,text filtering and other operations,the purpose is to organize the input text into the required text,this can reduce the expression match,punctuation,special The influence of symbols on the recognition effect,then the pre-processed short text is passed through the model to obtain the corresponding vectorized representation,then the feature vector is extracted through a singlelayer CNN,and finally the distribution probability of different intentions is output through the Softmax classifier,And then realize the recognition of user intent.(3)This thesis designs and implements an intelligent transportation customer service system based on the domain knowledge base on the basis of a thorough analysis of the research background and theoretical technology,and a WeChat public account as a platform.The system design scheme proposed in this thesis mainly includes user question analysis module,question answering module,and database management module.The user question analysis module is mainly to complete the domain classification,intention recognition,and entity recognition of the question.When the user’s question passes the question analysis module,it will help us determine the domain and intention of the question;the question and answer module can solve for traffic-related questions consulted by users,first of all,they are matched in the answer pairs of common questions.The similarity matching algorithm is used to find the answer.If the answer is found,the user is directly answered.If the system is not found,the question analysis template is used to determine the question.The entity in the sentence and the intention of the question,finally the system will find the matching answer in the domain knowledge base.Based on the analysis of the above technology and the proposed design scheme,this thesis completed the research and implementation of the intelligent transportation customer service system based on the domain knowledge base.Finally,the system passed the online test for a week,and the accuracy of the answer was higher than 85%.The results show that the system can run online to help the majority of driver users to solve the related consulting problems.